TomX8

I don't even know how to describe this properly. Basically, I hooked up an OBD2 reader to my RX8, exported the data and graphed it. The interior fans and AC have been switched off! It was also -10 outside when I measured this, so it's useless for the AC to do its work anyway.

This is my RX8 at a fixed throttle position at 1800rpm (ignore the value after 100K ms).


This graph shows the RPM's during the 1800rpm run where I kept the throttle at a fixed position. Notice how the RPM's NEGATIVELY correlate to the MAF g/s. As RPM's increase, MAF should increase aswell and vice versa. In this case, the Airflow increases, and RPM decreases, which doesn't make any sense whatsoever. Only when I pressed the throttle to get above 2000rpm (as you can clearly see), did the Airflow and RPM correlate to each other positively.


This is a graph of the idle values at rougly 800rpm.


The K values stand for the amount of seconds (20K ms = 20 seconds).

Notice how in the graph of 800rpm, the Short Term Fuel Trim jumps up and down radically.

I did some calculation, and it seems from this data that the MAF is the first to pick up a specific signal, to which the Trim values and O2 values follow. 1I stands for 1st interval Increase, and 1D stands for 1st interval Decrease. The values after are the miliseconds. when the increase or decrease in values occur. The MAF signal is High for 7 seconds, and low for 14 seconds, every time.


From this dataset I can conclude that the MAF is the first to pick up a change. The deviation betweens these interval values is so small and precise, that it suggests it's some signal causing this disturbance in MAF readings, or any other sensor connected to the ECU. A vacuum leak should be steady, and any malfunction like a dirty sensor or a misfire causing the sway in RPM and whatnot should be give irregular random values.

Some time before this, I have noticed that when the radiator fan turned itself on during idle, the short term fuel trims would drop, until the radiator turned off and they rose again. I will check this out tomorrow, and if this turns out to be related, it either means that a) the fan blows against something that causes the values to change (movement of air), or b) the electrical current of the fans affects the voltage of the MAF sensor cable. It probably doesn't make any sense, but it's the only think that I can think of.

Last weekend I have been moving around with my airbox, looking for a vacuum leak while my engine was running before I have found out about this data. When I moved the Airbox in a specific way, the LTFT and STFT would go towards 0% pretty rapidly, and it came back when I stopped fiddling. It was still attached to the airduct.

I have noticed something similar a year ago, when my radio would cut off for roughly 500ms. I decided to measure the interval, and it happened every 6 or 60 seconds, with a deviation of roughly 300ms (in which my thumb was the major factor).

So my actual question is, what can cause the interval in MAF readings? It works pretty much like connecting the MAF to an indicator light and turning it on. Is there something in a 2004 RX-8 Renesis HP which works according to 7-14 second intervals?

Loki

iTrader: (1)

The only thing I can think of is the A/C compressor. I know you said A/C was off, but it's still used if the cabin vents are in the full or partial "window" position. That, or my car's A/C controls don't always disengage for some reason. There is an audible click each time it engages, see if yours does that?

TomX8

I will definitely check for this. Last week I checked for vacuum leaks and heard a clicking noise in intervals, powering something on and off. Even though I turned off my AC, this could very well be it. Still, this is no reason for the airflow g/s value to increase while the RPM's decrease, unless the AC sucks its air through the engine intake, but I doubt anyone would like to smell gasoline and oil in their interior .

I added the RPM graph while you posted. It would makes sense for me if the ECU increases engine RPM to compensate for the additional load of the AirConditioning, and therefore, the airflow would rise too. But the airflow increases and RPM decreases, which doesn't make sense to me. How can an engine suck up more air while turning slower?

Loki

iTrader: (1)

This would make sense if there is increased load on the engine. More air = more work, but if all that work and then some goes into turning the A/C compressor, then your RPM still goes down.

Sort of like you can drive up a steep enough hill at full throttle and still loose speed.

Graph this with actual throttle position % and calculated load and see if that proves it.

TomX8

But the ECU still shouldn't need to compensate by increasing the Fuel Trim, because this messes up the Long-Term Fuel Trim as it already has done. My engine runs now idle on 13,3% LTFT, cruises with 8% LTFT. My car is starting with the 13,3% Long Term Fuel Trim as I have seen in my OBD2 readings. When I reset the ECU, the car starts much better, like it always did. Since it starts with 13,3% fuel trims, the engine starts but sputters and smokes. It also starts better when I deflood it. So the LTFT is floods the engine during starting. Hence why adding a bit of throttle helps, as it makes the mixture less rich, causing it to start within 3 seconds.


Imagine just adding a bit of voltage to the signal wire of the MAF. This would trick the ECU into thinking there is more air coming into the engine than it in reality is. The ECU injects the amount of fuel according to its Fuel Map (or however you call it), but concludes it isn't enough. So the ECU compensates by adding more fuel, which causes the surge in the STFT. Now add up that the ECU thinks that this mixture is correct, while in reality it injects more fuel than the engine actually needs and make the engine run too rich. A rich running engine has less power, and therefore should decrease in RPM.

It may be the very same case with the o2 sensor, since it gives the same signal interval, but the signal drops almost to 0 in these intervals. If the data would be finer and more precise, I believe it has hit the 0. A low voltage oxygen sensor screams there is much more oxygen than expected, and it should be corrected. Hence the ECU increases the Fuel Trim to compensate for it, then finds out that it's too much and lowers is to a negative fuel trim. But the average of the STFT evens out to a higher LTFT.


Hence why I believe that something is messing with the signal voltage of the MAF or the o2 sensor, because it really seems like it picks up voltage somewhere from something that turns on and off in the 7-14 seconds intervals.

I don't have the % Load values, but I do have the Throttle Position values. They are pretty much a straight line, except a minimal change down the road. Nothing that would drastically affect the RPM, nor cause the blocked signal of the MAF.

Loki

iTrader: (1)

At idle and low partial load it runs off the O2 sensor anyway, so the STFT is an output, not an input. Also as far as I know, modern ECUs make AFR, electronic throttle and open-loop fuel trim commands based on calculated load.

But, that's not to say that 13% idle LTFT is normal. It should be around 0.... vacuum leak?

The other thing is, usually sensor circuits shouldn't (with many asterisks) be super-sensitive to system-wide voltage fluctuations, since the sensor reading gets compared to a reference voltage. So if your sensor reads 3V on a 5V rail, which suddenly jumps to 5.2V (making the 3V jump to 3.1ish), the ECU should knows how to deal with that. Otherwise nothing would work.

TomX8

To summarize, during the surges of the MAF:
Throttle Position is Constant.
MAF increases from roughly 7g/s to 8g/s
RPM decreases.
o2 Sensor drops sharply to near 0.
Both MAF and o2 sensor are affected every 14 seconds.
The MAF for 7 seconds but flattens out (8g/s for entire 7 seconds).
The o2 sensor for 5 seconds, but peaks (keeps going down towards 0,2 and then climbs back to 0,8).

Note that I have another o2 sensor reading between 4500 and 5000rpm on the highway, where 4500 shows values between 0,8 and 0,5, and at 5000 the o2 sensor reads 0,1 during constant speeds and constant throttle on a straight and flat road. These readings are constant.

I suspect that something is activating every 14 seconds and presumably runs for roughly 7 seconds and messes up the sensor readings. I can't believe the o2 sensor or MAF are faulty, because the interval is just too precise. It behaves like an indicator light. Steady, precise, controlled.

Additionally, the 3 values in 1 graph for comparison. MAF reading, o2 voltage and a calculated AFR from o2 voltage. This suggest the MAF readings drop sooner than the o2 readings (as seen in the Excel screenshot).


Here is the STFT, o2 voltage and calculated AFR from o2 voltage. Here you can see how the STFT sits exactly on the readings of the o2 sensor. o2 goes down, STFT goes up right away.


I took the values of MAF, determined that all numbers above 7,65 are abnormal, and should be corrected with 0,8, which gave me a smooth realistic line. So it seems that the MAF sensor is pulling 0,8V from somewhere...


Correct me if I'm wrong. I am just presenting my findings and my interpretations. I'm not a car mechanic, and this is as far as I get. Tomorrow I will check whether I can find out what part of the car works according to this interval. When I find it, I will see if I can disconnect it and read the OBD2 again. I don't want to mask it by deactivating the source, but by eliminating the cause. This may be a better ground connection, wiring repair, or anything else that is necessary to do to keep all related parts functional.

EDIT: I just reminded myself that when I moved around with the Airbox while the car was running, the STFT and LTFT values changed, even as low as 1,3% for the LTFT and 0% for the STFT. By moving the airbox, I was also moving around the cable of the MAF sensor and changing pressure of the connector. I will check tomorrow by playing around with the cables and connector a bit, as well as trying to find the source of the interval.

TomX8

The STFT is indeed an output, since it is steered by the ECU to correct the fuel input. The ECU has a commanded AFR according to calculated load as you say, but the fuel trims are intended to bring the Actual AFR as close as possible to the Commanded AFR. So if the engine is running too lean according to readings, fuel will be added through trims. This does not mean that the Actual AFR is what is really is. It's what the sensors are reading, and if the readings are faulty, than the actual AFR is faulty aswell. Clearly something is going wrong, since the Idle has been trimmed to the extent that the car is flooding itself during the start, because it cranks the engine with the 13,3% fuel trim.

Most probably the MAF calculates according to the 5V rail Voltage and the voltage of the output, which gives a specific percentage of how much air is passing through, pretty much like you are saying. In that case, the value remains constant, but that's only my assumption. The fact is still that if you would add voltage to the signal wire, that is, if the ECU calculates according to the signal voltage, then the fluctuations of the 5V rail are irrelevant. If the 5V volt rail jumps to 5,5 and the signal reads 2,75, you know it's 50%. Same as a signal of 2,5v on a 5V rail would give off 50% aswell. But if you add a 0,5V to a 2,5V signal on a 5V rail, you end up with a value of 5/3=60%, which is 10% more, which is what is happening in my case.

The ECU must calculate according to the MAF, because it think it gets more air, adds fuel, but the RPM's drop becaus the engine runs too rich and loses power. But that's my logic.

The MAF seems to pick up some voltage from something that turns on and off in intervals. It could be something it shares its ground with too, and that this ground connection is just bad. I tried looking into the wiring diagram, but I can't find the ground locations anywhere...

Loki

iTrader: (1)

Another thing that might be on a regular 7 second cycle at idle is alternator voltage regulator, but I don't know much about them. It might be more of a ramp than a switch. It might also be related to rpm.

It could also be something silly like an O2 sensor heater.

I'm not sure about a wiring diagram with common grounds, but one thing that may help at least narrow it down is pulling all the accessory fuses, leaving only the bare minimum of what you need to run. That will at least tell you what it isn't.

Maybe I missed it, but if you are able to record data on a constant speed/constant load cruise, do you see the same fluctuations every 7 seconds?

TomX8

My battery died today. The engine cranked very slowly, until it just stopped turning and started clicking. Classic. You mentioned something like this in my other thread, Temporary Starting Problems. It may be that after a 200km drive, the battery got charged well enough, but any drive around 10-20km just didn't cut it because it charged poorly?

The alternator sounds plausible, given that it needs to recharge a dying battery constantly, and my car stood only still for 2 days (still in -15 degree Celsius tho). The battery is bought in May 2016, which makes it too young to actually die, but it happens, so I will send the battery to warranty. I will charge the battery overnight and see how the car behaves with a charged battery.

I doubt it would be an o2 sensor heater, since the heater is mainly there to warm the o2 sensor up during starts for improved emmissions? Or does the heater also work on idle?

Pulling fuses is a good idea, I will definitely check out which ones are needed for the car to keep running and give it a try.

Yes, I've did a constant load cruise (neutral standing still) at 2000rpm at a fixed throttle pedal position and it gave this 7 second fluctuation.

swoope

iTrader: (5)

also check maf connector for tightness.

beers

40w8

Doesn't A/C compressor run if you turn on defroster?

TomX8

I did, and it's tight .
I charged the battery and the car starts like new. Still need to try how it starts with these trims. First they got up to 13% for quite some time, and once the car got warm it got down to 7%.